Non-Associative Flux Algebra in String and M-theory from ... · Non-Associative Flux Algebra in String and M-theory from Octonions DIETER LÜST (LMU, MPI) Donnerstag, 27. Oktober

1

Supergravity @ 40, GGI Florence, October 27th, 2016

Non-Associative Flux Algebra in String and M-theory from Octonions

DIETER LÜST (LMU, MPI)

Donnerstag, 27. Oktober 16

1

Supergravity @ 40, GGI Florence, October 27th, 2016

In collaboration with M. Günaydin & E. Malek, arXiv:1607.06474

Non-Associative Flux Algebra in String and M-theory from Octonions

DIETER LÜST (LMU, MPI)


2

This talk is dedicated to my friend Ioannis Bakas


Outline:

3

II) Non-associative R-flux algebra for closed strings

I) Introduction

III) R-flux algebra from octonions

IV) M-theory up-lift of R-flux background

V) Non-associative R-flux algebra in M-theory


Point particles in classical Einstein gravity „see“ continuous Riemannian manifolds.

Geometry in general depends on, with what kind of objects you test it.

Strings may see space-time in a different way.

4

We expect the emergence of a new kind of stringy geometry.

I) Introduction

- [xi, x

j ] = 0


5

Closed strings in non-geometric R-flux backgrounds⇒ non-associative phase space algebra:

R. Blumenhagen, E. Plauschinn, arXiv:1010.1263.

D.L., arXiv:1010.1361;⇥x

i, x

j⇤

= i

l

3s

~ R

ijkpk

⇥x

i, p

j⇤

= i~�

ij,

⇥p

i, p

j⇤

= 0

=)⇥x

i, x

j, x

k⇤⌘ 1

3⇥⇥

x

1, x

2⇤, x

3⇤+ cycl. perm. = l

3sR

ijk


5



D.L., arXiv:1010.1361;⇥x

i, x

j⇤

= i

l

3s

~ R

ijkpk

⇥x

i, p

j⇤

= i~�

ij,

⇥p

i, p

j⇤

= 0

=)⇥x

i, x

j, x

k⇤⌘ 1

3⇥⇥

x

1, x

2⇤, x


3sR

ijk

This algebra can be derived from closed string CFT.R. Blumenhagen, A. Deser, D.L. , E. Plauschinn, F. Rennecke, arXiv:1106.0316

C. Condeescu, I. Florakis, D. L., arXiv:1202.6366

C. Blair, arXiv:1405.2283

D. Andriot, M. Larfors, D.L. , P. Patalong:arXiv:1211.6437

I. Bakas, D.L., arXiv:1505.04004


5



D.L., arXiv:1010.1361;⇥x

i, x

j⇤

= i

l

3s

~ R

ijkpk

⇥x

i, p

j⇤

= i~�

ij,

⇥p

i, p

j⇤

= 0

=)⇥x

i, x

j, x

k⇤⌘ 1

3⇥⇥

x

1, x

2⇤, x


3sR

ijk

This algebra can be derived from closed string CFT.R. Blumenhagen, A. Deser, D.L. , E. Plauschinn, F. Rennecke, arXiv:1106.0316

C. Condeescu, I. Florakis, D. L., arXiv:1202.6366

C. Blair, arXiv:1405.2283

D. Andriot, M. Larfors, D.L. , P. Patalong:arXiv:1211.6437


This algebra is also closely related to double field theory.R. Blumenhagen, M. Fuchs, F. Hassler, D.L. , R. Sun, arXiv:1312.0719


6

Two questions:


6

Two questions:

On the mathematical side:

How is the R-flux algebra related to other known non-associative algebras, in particular to the algebra of the octonions?


6

Two questions:



Can one lift the R-flux algebra of closed strings to M-theory?

On the physics side:


6

Two questions:



Can one lift the R-flux algebra of closed strings to M-theory?

On the physics side:

Our conjecture:the answers to these two questions are closely related


7

II) Non-geometric string flux backgrounds

Three-dimensional string flux backgrounds:

HijkTi�! f i

jkTj�! Qij

kTk�! Rijk , (i, j, k = 1, . . . , 3)

Chain of three T-duality transformations:

ds

2 = (dx

1)2 + (dx

2)2 + (dx

3)2 , B12 = Nx

3

H123 = NH-flux:

(Hellerman, McGreevy, Williams (2002); C. Hull (2004); Shelton, Taylor, Wecht (2005); Dabholkar, Hull, 2005)

(i) with H-flux:T 3


8

ds

2 =�dx

1 �Nx

3dx

2�2 + (dx

2)2 + (dx

3)2 , B2 = 0

Globally defined 1-forms:

⌘

1 = dx

1 �Nx

3dx

2, ⌘

2 = dx

2, ⌘

3 = dx

3

d⌘i = f ijk⌘j ^ ⌘k

Geometric flux: f123 = N

x

1(ii) Twisted torus tilde : T-duality along T 3

is a U(1) bundle over :T 3 T 2


9

(iii) Q-flux background:

ds

2 =(dx

1)2 + (dx

2)2

1 + N

2 (x3)2+ (dx

3)2 , B23 =Nx

3

1 + N

2 (x3)2

This background is globally not well defined, but it is patched together by a T-duality transformation.

C. Hull (2004)

T-duality along x

2

⇒ T - fold


10

(iii) Q-flux background:

ds

2 =(dx

1)2 + (dx

2)2

1 + N

2 (x3)2+ (dx

3)2 , B23 =Nx

3

1 + N

2 (x3)2

This background is globally not well defined, but it is patched together by T-duality transformation.

To make it well defined use double field theory:

Coordinates: (x1, x

2, x

3; x1, x2, x3)

C. Hull (2004)

W. Siegel (1993); C. Hull, B. Zwiebach (2009); C. Hull, O. Hohm, B. Zwiebach (2010,...)

SO(3,3) double field theory:

T-duality along x

2

⇒ T - fold


11

M. Grana, R. Minasian, M. Petrini, D. Waldram (2008); D. Andriot, O. Hohm, M. Larfors, D.L., P. Patalong (2011,2012);R. Blumenhagen, A. Deser, E. Plauschinn, F. Rennecke, C. Schmid (2013);D. Andriot, A. Betz (2013)

The dual background can then by described by „dual“ metric and a bi-vector:

Bij(x) ! �

ij(x) =12

⇣(g �B)�1 � (g + B)�1

⌘,

g(x) ! g(x) =12

⇣(g �B)�1 + (g + B)�1

⌘�1.

T ij

T ij

Qijk = @k�ij


11

M. Grana, R. Minasian, M. Petrini, D. Waldram (2008); D. Andriot, O. Hohm, M. Larfors, D.L., P. Patalong (2011,2012);R. Blumenhagen, A. Deser, E. Plauschinn, F. Rennecke, C. Schmid (2013);D. Andriot, A. Betz (2013)

The dual background can then by described by „dual“ metric and a bi-vector:

ds

2= (dx

1)2 + (dx

2)2 + (dx

3)2 , �

12 = Nx

3

Q-flux: Q123 = N

For the Q-flux background one obtains:

Bij(x) ! �

ij(x) =12

⇣(g �B)�1 � (g + B)�1

⌘,

g(x) ! g(x) =12

⇣(g �B)�1 + (g + B)�1

⌘�1.

T ij

T ij

Qijk = @k�ij


12

Then one obtains from the CFT of the Q-flux background the following commutation relation among the coordinates:

⇥x

1, x

2⇤

= Np

3

In general:

⇥x

i, x

j⇤

= i

l

2s

~

I

S1k

Q

ijk (x) dx

k = i

l

3s

~ Q

ijk p

k

winding number = dual momentum

Sigma-model for non-geometric backgrounds: A. Chatzistavrakidis, L. Jonke, O. Lechtenfeld, arXiv:1505.05457



13

(iv) R-flux background:

Buscher rule fails and one would get a background that is even locally not well defined.

T-duality along x

3


13

(iv) R-flux background:

Buscher rule fails and one would get a background that is even locally not well defined.

T-duality along x

3

R-flux can be defined in double field theory:

T k

�

ij(xk) ! �

ij(xk)

T k

x

k ! xk

Rijk = 3@[k�ij]


14

Strong constraint of DFT is violated by this background.

ds

2= (dx

1)2 + (dx

2)2 + (dx

3)2 , �

12 = Nx3

In our case we get:

R123 = NR-flux:

But it is still a consistent CFT background.


14


ds

2= (dx

1)2 + (dx

2)2 + (dx

3)2 , �

12 = Nx3

Now for the R-flux background we obtain:

⇥x

i, x

j⇤

= i

l

3s

~ R

ijkpk

⇥x

i, p

j⇤

= i~�

ij,

⇥p

i, p

j⇤

= 0

=)⇥x

i, x

j, x

k⇤⌘ 1

3⇥⇥

x

1, x

2⇤, x


3sR

ijk

In our case we get:

R123 = NR-flux:



14


ds

2= (dx

1)2 + (dx

2)2 + (dx

3)2 , �

12 = Nx3

Now for the R-flux background we obtain:

⇥x

i, x

j⇤

= i

l

3s

~ R

ijkpk

⇥x

i, p

j⇤

= i~�

ij,

⇥p

i, p

j⇤

= 0

=)⇥x

i, x

j, x

k⇤⌘ 1

3⇥⇥

x

1, x

2⇤, x


3sR

ijk

In our case we get:

R123 = NR-flux:

momentum



15

● Mathematical framework to describe non- geometric string backgrounds:

Group theory cohomology. ⇒ 3-cycles, 2-cochains, - products?

Two remarks:

● The same algebra appear in the context of the magnetic monopole.

R. Jackiw (1985); M. Günaydin, B. Zumino (1985)

I. Bakas, D.Lüst, arXiv:1309.3172D. Mylonas, P. Schupp, R.Szabo, arXiv:1207.0926, arXiv:1312.162, arXiv:1402.7306.



16

III) R-flux algebra from octonions

⌘ABC = 1() (ABC) = (123), (516), (624), (435), (471), (572), (673)

eAeB = ��AB + ⌘ABC eC

There exist four division algebras: over R , C , Q , O

Division algebra of real octonions : non-commutative, non-associative

O

Besides the identity, there are seven imaginary units eA

(A = 1 . . . , 7)


17

e1 e2

e3 e4e5

e6

e7

Remark: Octonions generate a simple Malcev algebraM. Günaydin, F. Gürsey (1973); M. Günaydin, D. Minic, arXiv:1304.0410.

Fano plane mnemonic:


18

e7

Split indices: e(i+3) = fi , for i = 1, 2, 3ei ,

and


18

e7

Split indices: e(i+3) = fi , for i = 1, 2, 3ei ,

and

[ei, ej ] = 2✏ijkek , [e7, ei] = 2fi ,

[fi, fj ] = �2✏ijkek , [e7, fi] = �2ei ,

[ei, fj ] = 2�ije7 � 2✏ijkfk

[ei, ej , fk] = 4✏ijke7 � 8�k[ifj] ,

[ei, fj , fk] = �8�i[jek] ,

[fi, fj , fk] = �4✏ijke7 ,

[ei, ej , e7] = �4✏ijkfk ,

[ei, fj , e7] = �4✏ijkek ,

[fi, fj , e7] = 4✏ijkfk

Associator [X,Y, Z] ⌘ (XY )Z �X(Y Z)


19

Contraction of octonionic Malcev algebra:

pi = �i�

12ei , x

i = i�

1/2

pN

2fi I = i�3/2

pN

2e7,


19

Contraction of octonionic Malcev algebra:

pi = �i�

12ei , x

i = i�

1/2

pN

2fi I = i�3/2

pN

2e7,

�! 0

[fi, fj ] = �2✏ijkek =) [xi, x

j ] = iN✏

ijkpk

[ei, ej ] = 2✏ijkek =) [pi, pj ] = 0

[fi, ej ] = ��

ije7 + ✏

ijkfk =)

⇥x

i, pj

⇤= i�

ijI

[xi, I] = 0 = [pi, I]

[fi, fj , fk] = �4✏ijke7 =)⇥x

i, x

j, x

k⇤

= N✏

ijkI

Agrees with non-associative R-flux algebra !


20

What is the role of un-contracted algebra?


20


Lift of R-flux algebra to non-geometric M-theory background:


20



- additional M-theory coordinatee7

⇒ Four coordinates: f1, f2, f3, e7


20





- but no additional momentum.

⇒ Three momenta: e1, e2, e3


20







⇒ Seven dimensional phase space !


20



Will be closely related to SL(4) /SO(4) exceptional field theory.





⇒ Seven dimensional phase space !


21

IV) M-theory up-lift of R-flux backgroundConsider IIA string: the duality chain splits into two

possible T-dualities:

HijkTij�! Qij

k , f ijk

Tjk�! Rijk


21



HijkTij�! Qij

k , f ijk

Tjk�! Rijk


21



HijkTij�! Qij

k , f ijk

Tjk�! Rijk

• two T-dualities ⇔ 3 U-dualities

(Need third duality along the M-theory circle to ensure right dilaton shift.)

Uplift to M-theory: add additional circle S1x

4

• 3-dim IIA flux background ⇔ 4-dim M-theory flux

background


22

(i) twisted torus T 3 ⇥ S1x

4

ds

24 =

�dx

1 �Nx

3dx

2�2 + (dx

2)2 + (dx

3)2 + (dx

4)2 , C3 = 0


22


4

ds

24 =

�dx

1 �Nx

3dx

2�2 + (dx

2)2 + (dx

3)2 + (dx

4)2 , C3 = 0

(ii) R-flux background: dualise along and . x

2, x

3x

4

This leads to a locally not well defined space.


22


4

ds

24 =

�dx

1 �Nx

3dx

2�2 + (dx

2)2 + (dx

3)2 + (dx

4)2 , C3 = 0


2, x

3x

4


Use SL(5) exceptional field theory:

10 generalized coordinates:

x

A $ x

[ab] (A = 1, . . . , 10 ; a, b = 1 . . . , 5)

x

↵ = x

5↵, (↵ = 1, . . . , 4)

D. Berman, M. Perry, arXiv:1008.1763

• 4 coordinates of :T 4

• 6 dual coordinates: (x41, x

42, x

43; x21, x

31, x

32)wrapped F1 wrapped D2


22


4

ds

24 =

�dx

1 �Nx

3dx

2�2 + (dx

2)2 + (dx

3)2 + (dx

4)2 , C3 = 0


2, x

3x

4




x

A $ x

[ab] (A = 1, . . . , 10 ; a, b = 1 . . . , 5)

x

↵ = x

5↵, (↵ = 1, . . . , 4)




42, x

43; x21, x

31, x


10 of SL(5)


22


4

ds

24 =

�dx

1 �Nx

3dx

2�2 + (dx

2)2 + (dx

3)2 + (dx

4)2 , C3 = 0


2, x

3x

4




x

A $ x

[ab] (A = 1, . . . , 10 ; a, b = 1 . . . , 5)

x

↵ = x

5↵, (↵ = 1, . . . , 4)




42, x

43; x21, x

31, x


10 of SL(5) 5 of SL(5)


22


4

ds

24 =

�dx

1 �Nx

3dx

2�2 + (dx

2)2 + (dx

3)2 + (dx

4)2 , C3 = 0


2, x

3x

4




x

A $ x

[ab] (A = 1, . . . , 10 ; a, b = 1 . . . , 5)

x

↵ = x

5↵, (↵ = 1, . . . , 4)




42, x

43; x21, x

31, x


10 of SL(5)

4 of SO(4)

5 of SL(5)


22


4

ds

24 =

�dx

1 �Nx

3dx

2�2 + (dx

2)2 + (dx

3)2 + (dx

4)2 , C3 = 0


2, x

3x

4




x

A $ x

[ab] (A = 1, . . . , 10 ; a, b = 1 . . . , 5)

x

↵ = x

5↵, (↵ = 1, . . . , 4)




42, x

43; x21, x

31, x


10 of SL(5)

4 of SO(4)

6 of SO(4)

5 of SL(5)


23

tri-vector:

SL(5) Flux-background:

g↵� =�1 + V 2

��1/3 ⇥�1 + V 2

�g↵� � V↵V�

⇤,

⌦↵�� =�1 + V 2

��1g↵⇢g��g��C⇢�� ,

ds2

7 =�1 + V 2

��1/3ds2

7 .

V ↵ =1

3!|e|✏↵��C��

@↵� = @↵� + ⌦↵��@� @

↵� =@

@x

↵�,

C. Blair, E. Malek, arXiv:1412.0635.

dual metric:

R↵,��⇢ = 4@↵[�⌦��⇢]R-flux:


23

tri-vector:

SL(5) Flux-background:

g↵� =�1 + V 2

��1/3 ⇥�1 + V 2

�g↵� � V↵V�

⇤,

⌦↵�� =�1 + V 2

��1g↵⇢g��g��C⇢�� ,

ds2

7 =�1 + V 2

��1/3ds2

7 .

V ↵ =1

3!|e|✏↵��C��

@↵� = @↵� + ⌦↵��@� @

↵� =@

@x

↵�,

C. Blair, E. Malek, arXiv:1412.0635.

dual metric:

R↵,��⇢ = 4@↵[�⌦��⇢]R-flux:

A particular choice of R-flux breaks SL(5) to SO(4).


24

In this way we obtain a well defined R-flux background in M-theory, which is dual to twisted torus:

R4,1234 = N

The R-flux breaks the section condition of exceptional field theory.

But it should be still a consistent M-theory background.

ds

2

7 = (dx

1)2 + (dx

2)2 + (dx

3)2 + (dx

4)2 , ⌦134 = Nx

24


25

What are the possible conjugate momenta (or windings)?

Four coordinates:x

1, x

2, x

3, x

4


25


Four coordinates:x

1, x

2, x

3, x

4

• Consider cohomology of twisted torus:

H1(T 3 ⇥ S1, R) = R3


25


Four coordinates:x

1, x

2, x

3, x

4

Dualize to IIB: H-flux with D3-branes

• Alternatively consider Freed-Witten anomaly:

R-Flux with momentum along

R-Flux with D0 branes.

x

4p4

⇕

This is forbidden by the Freed-Witten anomaly.

⇒ No momentum modes along the direction !x

4

• Consider cohomology of twisted torus:

H1(T 3 ⇥ S1, R) = R3


26

So we see that the phase space space of R-flux background in M-theory is seven-dimensional:

x

1, x

2, x

3, x

4 ; p1, p2, p3


26


x

1, x

2, x

3, x

4 ; p1, p2, p3

4 of SO(4) 3 of SO(4)


26


Missing momentum condition in covariant terms:

p↵R↵,��⇢ = 0

This condition is not the same as section condition.

x

1, x

2, x

3, x

4 ; p1, p2, p3

4 of SO(4) 3 of SO(4)


27


Identify

Xi =12ip

Nl3/2s �1/2fi , X4 =

12ip

Nl3/2s �3/2e7 , P i = �1

2i~�ei


27


Identify

Xi =12ip

Nl3/2s �1/2fi , X4 =

12ip

Nl3/2s �3/2e7 , P i = �1

2i~�ei

Octonionic algebra = conjectured M-theory algebra

[Pi, Pj ] = �i�~✏ijkP k ,⇥X4, Pi

⇤= i�2~Xi ,

⇥Xi, Xj

⇤= il3s

~ R4,ijk4Pk ,⇥X4, Xi

⇤= i�l3s

~ R4,1234P i ,⇥Xi, Pj

⇤= i~�i

jX4 + i�~✏i

jkXk ,⇥X↵, X� , X�

⇤= l3sR

4,↵��X� ,⇥Pi, X

j , Xk⇤

= 2�l3sR4,1234�[j

i P k] ,⇥P i, Xj , X4

⇤= �2l3sR

4,ijk4Pk ,

[Pi, Pj , Xk] = ��2~2✏ijkX4 + 2�~2�k[iXj] ,

[Pi, Pj , X4] = �3~2✏ijkXk ,

[Pi, Pj , Pk] = 0 .


28

Remarks:


28

Remarks:

gs : � / gs / R4 ⇒ String coupling

Natural identification of contraction parameter :�


28

Remarks:

The non-associative M-theory algebra is not SL(5) invariant.

However the algebra is SO(4) invariant:

Lie subalgebra [Pi, Pj ] = �i�~✏ijkP k

X↵: (2, 2) of SU(2, 2)

Pi : (3, 1) of SU(2, 2)




28

Remarks:

Further Modification compared to the string case:⇥Xi, Pj

⇤= i~�i

jX4 + i�~✏i

jkXk

The non-associative M-theory algebra is not SL(5) invariant.

However the algebra is SO(4) invariant:

Lie subalgebra [Pi, Pj ] = �i�~✏ijkP k

X↵: (2, 2) of SU(2, 2)

Pi : (3, 1) of SU(2, 2)




V) Outlook & open questions

29

Non-associative algebras occur in M-theory at many places:

- Multiple M2-brane theories and 3-algebras

I.Bakas, E. Floratos, A. Kehagias, hep-th/9810042

J. Bagger, N. Lambert (2007)

- backgroundsSpin(7), G2



29





What is the meaning of the odd-dimensional phase space?

Derivation from M-brane sigma modelSee also:D. Mylonas, P. Schupp, R.Szabo, arXiv:1312.1621




29


Generalization to higher dimensional exceptional field theory?




What is the meaning of the odd-dimensional phase space?

Derivation from M-brane sigma modelSee also:D. Mylonas, P. Schupp, R.Szabo, arXiv:1312.1621



Documents

Non-Associative Flux Algebra in String and M-theory from ... · Non-Associative Flux Algebra in String and M-theory from Octonions DIETER LÜST (LMU, MPI) Donnerstag, 27. Oktober